多天线频谱感知与接入模型协议研究

【作者】 鲁进；

【导师】 黄铭；

【作者基本信息】 云南大学 ， 通信与信息系统， 2014， 博士

【摘要】 无线技术由于可以灵活方便地接入通信网络而备受青睐。据Erikson全球移动公司报道,2013年手机用户数达到了75亿,这一数字已超过当前70亿的世界人口数。无线通信已成为服务技术的主流且规模正在不断扩大,人们不得不在公用网内开发云端技术来处理大数据量的传输。随着网络用户向移动互联网的转移和物联网技术的发展,无线传输的容量和可靠性被赋予了更高的要求。然而如何提高频谱利用率、采用有限的频谱资源承载这些业务是无线通信技术进一步发展必须要解决的首要问题。固定的频谱分配政策是造成频谱利用率低下的主要原因。2003年美国联邦通信委员会(FCC)和国际电信联盟(ITU)等频谱机构发现各地区的频谱利用率非常低下。为了提高认知系统的频谱使用效率和对频谱环境的适应能力,无线网络需要具有传感能力。1999年美国提出了通过谱传感利用频谱的认知无线电标准,采取动态频谱管理技术允许非授权用户伺机接入授权用户的空闲频谱而实现对频谱的有效利用。据称该技术可以完全解决频谱匮乏的问题,已被认为是未来无线网络的核心技术之一。动态利用频谱需要解决频谱感知和频谱接入两个方面的问题。前者要求系统对空闲频谱实施快速而准确的检测；后者要求用户能自适应地接入频谱空洞,并在发现主用户的情况下快速地切换到其他信道。隐终端和无线衰落等不利的环境因素使得用户在相互覆盖范围内无法发现潜在的通信冲突,使得感知性能大打折扣；同时当网络负载较大时,频谱共享系统的服务质量会下降。本文对这两个方面涉及的理论与技术问题进行深入探索,主要工作如下：首先,介绍认知无线电的研究背景,说明当前频谱危机和授权频谱利用率低下的现实。然后对认知无线电技术的概念、特点、各种标准、项目和应用前景进行了阐述。第二,研究了多天线分集合并技术和频谱感知技术,并提出一种多天线的智能谱感知方法,该方法融合了匹配滤波器检测、能量检测和循环平稳检测算法。通过假设信道存在高斯噪声和平坦Nakagami衰落而建立了相关的二元假设检验模型,并推导出瞬时检测概率的封闭表达式；根据SC、EGC、MRC三种多天线合并方式下的幅度增益的随机统计特性,分别解析了瞬时信噪比对应的概率分布。通过将瞬时检测概率在其概率密度函数上进行平均而导出精确的或近似的平均检测概率表达式,同时根据各检测算法的事件概率计算出平均检测时间的表达式。其后利用MATLAB的数值计算工具分析了检测概率和检测时间的ROC特性。第三,研究动态频谱接入的协议和模型。提出了一种基于服务期损失制轮询调度的有中心混合频谱接入方案,次用户根据感知结果和干扰温度判断主用户的活动状态,并概率式地接入信道。通过构建Nakagami信道下的MAC层排队模型,解析了次用户的平均服务率的封闭表达式。为了保证主用户的服务质量,规定主用户具有最大容忍延时,以此为约束条件分析了平均服务率的理论最大值。由于次用户的业务类型往往多样化,该部分又进一步研究了次用户系统具有多队列排队结构的情形,考虑了两种调度方案：第一种为概率式调度方案,采用拉格朗日乘子法得到了各个队列的接入概率；第二种是损失制服务轮询调度机制,在传统的门限服务、限定1服务和完全／门限两级优先级服务三种的轮询模型中假设有顾客离开,然后采用嵌入式Markov链和概率母函数的分析方法建立了与此调度方案相吻合的数学模型及其函数关系式,并精确解析出系统特性参数(平均排队队长、平均循环周期、平均等待时延)的表达式。其后利用MATLAB的数值计算工具分析了以上参数的ROC特性。第四,基于OPNET软件,根据前节的多天线感知和轮询频谱接入方案设计了认知无线电感知接入仿真平台,包含三个方面：在物理层,用Simulink搭建信号衰落、加性高斯信道、产生PU信号类型等过程；在MAC层,用OPNET搭建节点模块、数据源产生模块、队列模块；物理层和MAC层通过OPNET和MATLAB的联合仿真引擎接口进行信息交互。然后利用该实验平台对设定的仿真场景下的频谱利用率、平均排队队长和平均等待时延等相关参数进行了测试研究。更多还原

【Abstract】 Wireless technology is favored since it could access communication networks flexibly and easily. According to the report of global mobile company Erikson, the number of mobile phone users has reached7.5billion in2013, which has exceeded the current world population of7billion. Wireless communication has become the mainstream of service technology and its scale is expanding continuously, and thus people have to develop cloud technology in the public network to handle the large amount of data transmission. With the shift of network users to mobile Internet and the development of the technology of Net of Things, the higher demands about capacity and reliability of wireless transmission were put forward. Therefore, it is the primary issue of how to improve the spectrum efficiency and use limited spectrum resource to carry these businesses as the further development of wireless communication technology.Fixed spectrum allocation policy is the chief cause of low spectrum efficiency. In2003, Federal Communications Commission (FCC), International Telecommunication Union (ITU) and other spectrum institutions found that the spectral efficiency for all regions was very low. In order to improve spectral efficiency and the adaptability in spectrum environment of cognitive system, wireless networks need to have the capability of sensing. In1999, the United States proposed the Cognitive Radio standard of using spectrum through spectrum sensing, which allowed the unauthorized user to access to idle spectrum of authorized user opportunistically using dynamic management techniques, and to use the spectrum effectively. It is reported that the problem of spectrum shortage could be completely solved, and Cognitive Radio is considered to be one of the core technologies in future wireless network.To use spectrum dynamically, the problem of spectrum sensing and spectrum access should be solved. The former requires the system implementing rapid and accurate detection to idle spectrum; the latter requires users accessing spectrum holes adaptively, and switching to another channel quickly when the primary user was found. Some adverse environmental factors, such as hidden terminal and wireless fading, will make users can not find potential communicate conflicts in the mutual coverage, and make wireless sensing performance reduce greatly; Meanwhile, the quality of service of spectrum sharing system will decline when the network load becomes large. This dissertation deeply explored the theory and technical issues involved in the above two aspects. The main work is listed as follows:First of all, the research background of Cognitive Radio was introduced, which indicated that the reality of the current spectrum crisis and low utilization rate of authorized spectrum. Then the concept, features, standards, projects and prospects of Cognitive Radio technology were described.Second of all, the multi-antenna diversity combining technology and spectrum sensing technology were studied, and a multi-antenna intelligent spectrum sensing method was proposed, which combined the Matched Filter Detection, Energy Detection and Cyclostationary Detection algorithms. The relevant Binary Hypothesis Testing model was established through assuming the presence of Gaussian noise and flat Nakagami fading channel, and the closed expressions of instantaneous detection probabilities were deduced; According to the random statistical properties of amplitude gain under the three multi-antenna combine modes of SC, EGC, MRC, the probability distribution characteristic of instantaneous signal to noise ratio was analyzed separately. By averaging the instantaneous detect probability on the probability density function of instantaneous signal to noise ratio, the exact or approximate average detection probability expression was derived, and the expression of average detection time was also obtained according to the event Probability of every detection algorithm. Subsequently, the ROC characteristics of detection probability and detection time were analyzed by MATLAB numerical calculation tool.Third of all, the protocol and model of Dynamic Spectrum Access were studied. A centred and mixed spectrum access scheme based on losing service polling scheduling was proposed, in which Second Users judge the active state of primary user according to the results of sensing and interfere temperature and then access channel by a probability. By constructing the MAC layer queuing model under Nakagami channel, the closed expression of average service rate of Second User was analyzed. In order to ensure the quality of service for Primary Users, we specified the Primary User who has the maximum tolerable delay, in which the theoretical maximum Average Service Rate was analyzed. Because the business type of Second User was diverse, the case of multi-queue structure in second user system was further investigated, in which two scheduling schemes was considered:the first one was probabilistic scheduling scheme, and the access probability of each queue was obtained according to Lagrange Multiplier Method; the second one was the losing service polling scheduling mechanism, in which we supposed that the customer will leave in the three traditional polling models of Gated service, Limit1service and Exhaustive-Gated two levels priority service. The mathematical model and function relationship was established using the analytical method of embedded Markov chain and Probability Generating Function, and the expression of system parameters, such as Average Queue Length, Average Cycle Time, Average Waiting Time, were precisely parse out. Subsequently, the ROC characteristics above were analyzed by MATLAB numerical calculation tool.Finally, based on OPNET software, Cognitive Radio sensing and access simulation platform was designed according to multi-antenna spectrum sensing and polling spectrum access scheme above, which included three aspects:in physical layer, the process of signal fading, additive Gaussian channel, generate primary user signal types were build with Simulink; in MAC layer, node module, data sources generating module and queue module were build with OPNET. Physical layer and MAC layer exchange information through OPNET and MATLAB co-simulation engine interface. Then we used the experimental platform to test and research the relevant parameters by setting a simulation scenario, including spectrum utilization, Average Queue Length and Average Waiting Time.更多还原